资源简介
稀疏表示算法(Sparse Representation Classification,SRC)一种广泛应用于人脸识别的算法。
代码片段和文件信息
function testClassPredicted=bootstrapnnlsClassifier(trainSettrainClasstestSettestClassoption)
% Bootstrap NNLS Classifier: testSet=trainSet*Y s.t. Y>=0.
% Usage:
% [testClassPredictedsparsity]=bootstrapnnlsClassifier(trainSettrainClass[]testClass)
% [testClassPredictedsparsity]=bootstrapnnlsClassifier(trainSettrainClasstestSettestClass)
% [testClassPredictedsparsity]=bootstrapnnlsClassifier(trainSettrainClasstestSettestClassoption)
% trainSet matrix the training set with samples in columns and features in rows.
% trainClass: column vector of numbers or string the class labels of the traning set.
% testSet: matrix the test set.
% testClass: column vector of numbers or string the class labels of the
% test/unknown set. It is actually unused in this function thus set it [].
% option: struct the options to configue this function:
% option.method string the optimization algorithm used to solve the NNLS problem. It could be
% ‘nnls‘: used the NNLS algorithm (default);
% ‘seminmfupdaterule‘: use the update rules based algorithm;
% ‘sparsennls‘: used NNLS algorithm with sparse constraint.
% option.predicter: the method to find the class label of a test sample according to Y. It could be
% ‘max‘: the same class label with the training sample with the maximum coefficient (default);
% ‘kvote‘: select k training samples with the k largest coefficients and decide the class labels by majority voting.
% option.kernel string specifies the kernel. can be ‘linear‘(default)‘polynomial‘‘rbf‘‘sigmoid‘‘ds‘
% option.param scalar or column vector the parameters for kernels the default is [].
% option.kernelParamRandomAssign: logical if randomly assign the
% parameters the default is false.
% option.k: scalar only for option.predicter=‘kvote‘. The default is 1.
% option.numRandom scalar the times to use bootstrapping. The default is 99.
% testClassPredicted: column vector the predicted class labels of the test/unknown samples.
% sparsity: scalar the sparsity of the coefficient matrix Y.
% References:
% [1]\bibitem{nips2011}
% Y. Li and A. Nogm
% ‘‘Non-neagtive least squares classifier‘‘
% {\it Advances in Neural Information Processing Systems}
% submitted.
% Available at \url{http://cs.uwindsor.ca/~li11112c/doc/nips2011.pdf}
% Contact Information:
% Yifeng Li
% University of Windsor
% li11112c@uwindsor.ca; yifeng.li.cn@gmail.com
% May 23 2011
if nargin<5
option=[];
end
optionDefault.method=‘nnls‘;
optionDefault.predicter=‘max‘;
optionDefault.kernel=‘linear‘;
optionDefault.param=[];
optionDefault.kernelParamRandomAssign=false;
optionDefault.k=1;
optionDefault.numRandom=99;
option=mergeOption(optionoptionDefault);
trainSetOrigin=trainSet;
trainClassOrigin=trainClass;
testSetOrigin=testSet;
if size(trainSetOrigin3)>1 % tensor
trainSetOrigin=matrizicing(trainSetOrigin3);
testSet=matrizicing(testSetOrigin3
属性 大小 日期 时间 名称
----------- --------- ---------- ----- ----
目录 0 2015-03-07 07:27 srv1_9\
文件 3508 2015-03-02 21:56 srv1_9\readMe.txt
文件 509 2015-03-02 20:47 srv1_9\exampleUSR.m
文件 858 2013-02-14 20:23 srv1_9\wvote.m
文件 7186 2015-03-07 07:10 srv1_9\vsmf.m
文件 821 2011-10-11 17:00 srv1_9\vote.m
文件 290 2010-09-28 04:54 srv1_9\vec2mat.m
文件 4308 2013-02-23 00:12 srv1_9\usr.m
文件 1397 2011-10-11 17:10 srv1_9\unmatrizicing.m
文件 4116 2013-02-04 09:04 srv1_9\threeDSearchUniverse.m
文件 1975 2013-07-11 02:27 srv1_9\subspace.m
文件 3170 2015-03-02 18:45 srv1_9\SRC2.m
文件 3150 2011-12-17 02:32 srv1_9\src.m
文件 372 2013-07-11 00:07 srv1_9\sparsity.m
文件 1450 2012-01-12 02:46 srv1_9\sparsenmfnnlstest.m
文件 7046 2012-11-11 21:25 srv1_9\softSVMTrain2.m
文件 2178 2012-09-15 23:53 srv1_9\softSVMPredict2.m
文件 3206 2012-11-25 08:53 srv1_9\significantAcc.m
文件 961 2011-11-19 03:51 srv1_9\sampleSelNNLS.m
文件 493 2011-11-19 04:03 srv1_9\sampleSelKNN.m
文件 570 2011-10-11 16:43 srv1_9\pseudoinverse.m
文件 486 2013-04-04 06:19 srv1_9\proximalOperator.m
文件 399 2012-04-04 22:35 srv1_9\plotTime.m
文件 1795 2012-11-25 08:59 srv1_9\plotNemenyiTest.m
文件 1808 2012-04-05 07:03 srv1_9\plotDataMulti.m
文件 2525 2013-02-19 22:13 srv1_9\plotBarError.m
文件 4100 2012-08-10 00:12 srv1_9\perform.m
文件 869 2011-10-11 17:14 srv1_9\normmean0std1.m
文件 175 2011-11-19 03:35 srv1_9\normcl1.m
文件 393 2012-02-17 09:22 srv1_9\normalizeKernelMatrix.m
文件 248 2013-05-22 23:09 srv1_9\NNQPSMOMulti.m
............此处省略100个文件信息
- 上一篇:Matlab揭秘
- 下一篇:用卷积神经网络实现彩色图像的超分辨率matlab
评论
共有 条评论